www.nature.com/scientificreports

OPEN polymyxa bioflm polysaccharides antagonise Fusarium graminearum Received: 31 July 2018 Salme Timmusk1, Dana Copolovici2, Lucian Copolovici2, Tiiu Teder1,3, Eviatar Nevo4,5 & Accepted: 12 December 2018 Lawrence Behers1,6 Published: xx xx xxxx Fusarium Head Blight (FHB) caused by Fusarium graminearum pathogens constitutes a major threat to agricultural production because it frequently reduces the yield and quality of the crop. The disease severity is predicted to increase in various regions owing to climate change. Integrated management where biocontrol plays an important role has been suggested in order to fght FHB. P. polymyxa A26 is known to be an efective antagonist against F. graminearum. Deeper understanding of the mode of action of P. polymyxa A26 is needed to develop strategies for its application under natural settings in order to efectively overcome the pathogenic efects. This study aims to re-evaluate a former study and reveal whether compounds other than non-ribosomal lipopeptides could be responsible for the antagonistic efect, despite what is often reported. Wheat seedlings were grown to maturity and the spikes infected with the pathogen under greenhouse conditions. The development of FHB infection, quantifed via the disease incidence severity and 100-kernel weight, was strongly correlated (r > 0.78, p < 0.01) with the content of the polysaccharide component D-glucuronic acid in the bioflm. Furthermore, while increased inoculum density from 106 to 108 cells/ml did not afect wild type performance, a signifcant increase was observed with the P. polymyxa mutant defcient in nonribosomal lipopeptide synthesis. Our results show that P. polymyxa A26 bioflm extracellular polysaccharides are capable of antagonizing F. graminearum and that the uronate content of the polysaccharides is of critical importance in the antagonism.

Fusarium head blight (FHB) is a devastating disease of cereals and is caused by a group of pathogens of which F. graminearum prevails in Nordic countries. Te disease severity is predicted to increase in various regions own- ing to climate change1. FHB is of particular concern because many of the Fusarium species produce mycotoxins that contaminate infected grain and may pose a serious threat to human and domestic animal health. Grain that has been infected with the fungus may become incorporated into our staple diets. For these reasons the antifun- gal activity of biocontrol agents against FHB pathogens has been extensively studied and several mechanisms suggested2–5. Paenibacillus polymyxa is known as an FHB biocontrol agent. It is generally recognised that surface-associated colonise as bioflms, which are microniches entirely diferent from their surroundings. Tis allows the bacteria to work as a functional unit, accomplishing tasks not possible for their planktonic state. Bioflms consist of cells and matrices where complex exopolysaccharides and proteins are major components, and they can pro- vide an important bacterial survival strategy in natural systems6–8. P. polymyxa biocontrol ability is ofen linked to the large bacterial pool of bioactive compounds such as nonribosomal peptides/polyketides (NRPs/PKs)4,9. Despite enormous diversity the compounds have common regulatory systems, as they are produced nonriboso- mally and require activation by phosphopantetheinyl transferases (PPTase) of which Sfp-type PPTase is required for activation of peptidyl and acyl carrier domains. Te broad range of activities of NRPs and PKs includes pro- motion of adaptation to unfavourable environments. Te P. polymyxa A26 strain was isolated from the Evolution Canyon, South Facing Slope, Israel (EC SFS)10,11 (Fig. 1 and Table 1) and has co-evolved with wild progenitors of

1Department of Forest Mycology and Plant Pathology, Swedish University of Agricultural Sciences, Uppsala, Sweden. 2Faculty of Food Engineering, Arad University of Aurel Vlaicu, Arad, Romania. 3Department of Plant Physiology, Estonian University of Life Sciences, Tartu, Estonia. 4International Graduate Centre of Evolution, University of Haifa, Haifa, Israel. 5National Academy of Sciences, Washington, USA. 6Nova West Technologies & Communications, Tucson, AZ, USA. Correspondence and requests for materials should be addressed to S.T. (email: salme.timmusk@ slu.se)

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 1 www.nature.com/scientificreports/

Figure 1. Te Evolution Canyon (EC) model. (A) Schematic diagram. (B) Cross section view of EC at Lower Nahal Oren, Mount Carmel. (C) Air view of EC (source10: Nevo, 2012 Evolution Canyon,” a potential microscale monitor of global warming across life, PNAS 109; 8) (Photo by E. Nevo).

Name and abbreviation Origin Publications Paenibacillus polymyxa A26 (A26) Wild barley rhizosphere, Evolution Canyon, Haifa, Israel Timmusk et al.11 P. polymyxa A26∆sfp (A26Sfp) Wild barley rhizosphere, Evolution Canyon, Haifa, Israel Timmusk, S. et al.9 Fusarium graminearum A602/1998 Te National Veterinary Institute, Sweden Abd El-Daim et al.12

Table 1. Strains and primers used in the study.

cereals over a long period of time, helping host plants to adapt and survive under diverse harsh conditions11. Te genome analysis reveals regions that refect these adaptations and the isolate is a superior plant growth promoting bacterium (PGPB) compared to isolates from more moderate environments11. In order to study the performance of the isolate we inactivated its Sfp-type PPTase gene and created a mutant that is incapable of producing enzy- matically active 4′-phosphopantetheinyl transferase. Tis in turn results in a strain lacking enzymatically active NRPs and PKs (A26∆sfp), and the nonribosomal peptide and polyketides, ofen reported as active metabolites for the biocontrol4,9, are not produced. Formerly we established a screening method for developing FHB biocontrol agents based on combining dual plate assays with a kernel assay12. Using the method, dual plate assays confrmed that NRPS/PKs products have critical importance for the antagonistic activity of A26. Studies with the more complex system employing a wheat grain assay showed, however, that in the case of one FHB-causing pathogen, F. culmorum, bioflm matrix for- mation may be the biocontrol agent of major importance in antagonising the pathogen. At the same time, only limited antagonism of the other pathogen, F. graminearum, was observed in the wheat grain assay9,12. Tis raised the question of why, in the case of F. graminearum, bioflm antagonism did not work and only the NRPS/PKs products were of critical importance. Te present study was conducted to re-evaluate the role of the P. polymyxa A26 extracellular matrix in antag- onism to F. graminearum. Te mutant (A26∆sfp), which lacks the ability for NRP and PK synthesis9, was studied under two inoculum densities, low (106 cells/ml) and high (108 cells/ml). We show that the mutant has compara- ble antagonistic ability to that of the wild type at the high inoculum density, and that the ability is correlated with the content of uronates in the bioflm polysaccharides. While many studies have been performed aiming to reveal the mechanism of antagonism of the economically important FHB pathogens, to the best of our knowledge this is the frst report on the involvement of the uronate components of the bioflm polysaccharides. Material and Methods Experimental set-up. Tree experimental series were performed:

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 2 www.nature.com/scientificreports/

1. Te wheat grain assay was performed at two inoculum densities in order to re-evaluate the role of the A26 extracellular matrix in antagonizing F. graminearum. A26∆sfp was applied at low and high inoculum densities (106 and 108 cells/ml respectively). Te results clearly showed the superior efect using the higher inoculum density. 2. Developing biocontrol agents requires combining screening strategies. Hence, greenhouse experiments on wheat plants grown to maturity were performed to test the efect in more natural settings. Wheat head assays evaluating disease severity (DS), disease incidence (DI) and 100-kernel weight (100-kw) were performed. We were further interested in the involvement of exopolysaccharides (EPS), which are the largest group of bioflm components, and along with wheat head assays their bioflm EPS assessment was performed. 3. In order to confrm the role of EPS in disease suppression, A26∆sfp and A26 EPS assays were performed on wheat kernels along with assays of supplied levan and alginate. Commercial samples of levan and sodi- um alginate were used as reference substrates owing to their contrasting uronate contents (Table S1).

Microbial growth and culture conditions. P. polymyxa A26 was isolated from the South Facing Slope at the natural laboratory called the Evolution Canyon, Israel (Table 1). Te Sfp-type 4-phosphopantetheinyl trans- ferase deletion mutant strain was previously generated as described by us earlier9. Stock cultures were stored at −80 °C and were streaked for purity on half strength tryptic soy agar (1/2 TSA, pH 6.2). 10 ml of half strength tryptic soy broth (1/2 TSB, pH 6.2) was inoculated with single cells and grown at 30 ± 2 °C for 12 h. 100 µl of the preinoculum with 108 cells per ml was used for 100 ml fask culture inoculations. Bacterial strains were grown in 1/2 TSB, pH 6.2, 180 rpm at 30 ± 2 °C, for 72 h. Cultures were centrifuged and pellets resuspended in sterile water as described earlier12. Finally, the cultures were adjusted to 106 and 108 cells/ml. Te fungal pathogen F. graminearum strain A602/1998 was obtained from the National Veterinary Institute, Uppsala, Sweden and was previously characterised for virulence12. Te pathogen was grown on potato dextrose agar (PDA) plates for seven days at 22 °C. Macroconidia for inoculation were obtained by fooding the surface of colonized agar with sterile or phosphate bufered saline (PBS). Te resulting inoculum density of the suspension was 105/ml.

Plant treatment. Winter wheat (Stava) seeds were surface sterilised by a 60 s wash in 99% ethanol, followed by a 6 min wash in 3% sodium hypochlorite solution, a wash in 99% ethanol, and repeated rinsing in sterile water. Te seedlings were grown in greenhouse soil in 30 cm diameter pots. Afer a month of growth, the plants were vernalized for two months. Troughout the growing season the plants were watered daily with a standard nutritional solution. Te experiment was performed in four replicates, each consisting of 20 plants. Wheat heads were sprayed with 100 µl of bacterial solutions at either low or high inoculum density (106 and 108 cells per ml respectively) at the beginning of the fowering stage (BBCH 61). Te control plant heads were treated with 100 µl of sterile water. One week afer the antagonist treatment (end of fowering stage) 30 µl of macroconidia suspen- sion was used to inoculate a single central foret on each wheat head (four replicates each containing twenty plants). Controls were heads treated only with water or with pathogen suspension. Inoculated spikes were misted with water and covered with plastic bags for 10 days. Immediately before the pathogen inoculation, A26 and A26∆sfp bioflms from twenty wheat heads from each treatment were carefully fushed into 10 ml sterile water in 50 ml Falcon tubes for the bioflm studies. As controls, untreated wheat heads were fushed with sterile water. Counting of bacterial colony forming units was performed on ½ TSB plates. Te bioflm bacteria were re-isolated by selective plating and confrmed by PCR as described by us earlier13. Wheat heads were collected 21 days afer inoculation at the fully ripe stage (BBCH 89) and scored for disease incidence and severity on a scale from 0 to 100%. Disease severity was evaluated based on visual assessment of heads exhibiting FHB disease symptoms. F. graminearum presence was randomly confrmed using PCR12 and fuorometric assay. Afer the FHB evaluation the heads were allowed to dry and 100-kernel weights were determined.

Wheat head assessment, and polysaccharide and glucuronate content evaluation. Wheat heads were fushed with 10 ml deionised sterile water immediately before the pathogen treatment in order to collect the bioflms developed afer the A26 and A26∆sfp inoculations. Te bioflm bacteria were re-isolated by selective plating, confrmed by PCR as described9,13, quantifed, the supernatant polysaccharides isolated and D-glucuronate (D-GA) contents recorded. Te bioflm samples in water were used for polysaccharide isolation as described earlier14. Te pellet was lyophilised, weighed and stored at −4 °C until uronic acid content was assessed. Te assay was based on hydrolysis of glucoside bonds that bind the polysaccharides. Tis step digests the poly- saccharides into their component monosaccharides. Analysis of uronic acid content was performed as described earlier by Mojica et al.15 with small modifcations. Briefy, the bioflm pellets were weighed and dissolved in 200 µl of deionised water. Potassium sulfamic acid (4 M, pH 1.6) was added to the bioflm solution and mixed by vor- texing. Ten sodium tetraborate solution in concentrated sulfuric acid (0.0125 M) was added. Te solutions were incubated for 5 min in a 100 °C water bath, cooled on ice for 3 min and centrifuged at 2,000 rpm for 10 min afer which 20 µl hydroxyphenol solution (0.15% v/v) was added to the supernatant. Te solution was then mixed gen- tly and the absorbance read at 520 nm. Each data point represents the average of twenty replicate measurements. P. polymyxa A26∆sfp cell and polysaccharide assay on wheat grains. The experiments were performed as reported earlier12. Briefy, conical fasks containing 20 g wheat grains were inoculated with 15 ml A26∆sfp at 106 or 108 cells/ml. For the polysaccharide assay, the kernels were treated with 15 ml of A26∆sfp and A 26 polysaccharide solution with the titres 10 µg/ml. Cells were grown and polysaccharides isolated as indicated below. Controls were treated with sterile water. In order to study the efect of uronates, two reference polymers

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 3 www.nature.com/scientificreports/

contrasting in uronate content were applied: alginate (Sigma) and levan (Sigma). Te polysaccharide powders were dissolved in deionized sterile water to the fnal concentration 15 µg/ml. 20 g wheat grains were mixed with 15 ml of the polysaccharide solution. Flasks were incubated at room temperature for 8 h, and then inoculated with 1 cm2 agar plugs from 2 week old cultures of F. graminearum.

EPS extraction. EPS extraction was performed as described earlier with small modifcations14. Briefy, bac- terial cultures were diluted 1:5 with distilled water and centrifuged for 30 min at 17,600 g at 20 °C to separate cells. Ten, EPS were precipitated by slowly pouring the supernatant into two volumes of isopropanol while stirring at 200 rpm. Te fltered polysaccharide was suspended in a digestion solution consisting of 0.1 M MgCl2, 0.1 mg/ml DNase, and 0.1 mg/ml RNase solution, and incubated for 4 h at 37 °C. Samples were extracted twice with phenol-chloroform, lyophilised using a Virtis SP Scientifc 2.0 freeze dryer, taken to the initial volume and dialysed against distilled water.

DNA extraction and quantification. The grain samples were freeze-dried and ground into a fine powder using a Precellys 24 homogenizer (Bertin Technologies, France). Samples were lysed by incubating 100 mg powder for 15 min in 350 μl of glucose bufer as described by us earlier16. DNA was extracted using a hexadecyl-trimethyl-ammonium bromide-based method17. Te bacterial strains were confrmed by PCR as described by us earlier13. Fungal growth was assessed visually and randomly verifed using PCR and sequencing afer 10 days of growth12. For F. graminearum PCR ITS1F and ITS4 primers18 (5′-CTT GGT CAT TTA GAG GAA GTAA-3′ and 5′-TCC TCC GCT TAT TGA TAT GC-3′) were used. Initial denaturation at 95 °C was fol- lowed by 35 amplifcation cycles: denaturation at 95 °C for 30 sec, annealing at 58 °C for 30 sec, and extension at 72 °C for 30 sec, followed by an extension step of 72 °C for 7 min. Te reaction mix contained 200 µm dNTP, 2.75 mM MgCl2, 0.025 U/µl polymerase (DreamTaq Green, Termo Scientifc, Waltham, MA, USA) and 0.2 µM of each primer. For the pathogen quantifcation the dsDNA of the PCR mix was purifed using a Nucleo Spin kit (Macherey-Nagel, PA, USA) and quantifed using the Invitrogen Qubit 4 Fluorometer according to the manufacturer’s instructions. ™ ™

Qualitative determination of bioflm composition by UHPLC-MS+. Te chromatographic analyses have been performed using a liquid chromatograph (Nexera X2, Shimadzu, Tokyo, Japan) equipped with a diode array detector (M30A, Shimadzu, Tokyo, Japan) and a mass spectrometer (Model 8040, Shimadzu, Tokyo, Japan). Te separation of compounds was performed on Nucleodur 100-5-NH2-RP column (4.6 mm i.d. × 250 mm col- umn length, 5 µm particle size, Macherey-Nagel GmbH, Duren, Germany). Te column temperature was main- tained at 35 °C and the fow rate at 1 ml min−1. Te solvents used for the chromatographic elution consisted of ultra-pure water with 0.1% TFA (A) and acetonitrile (B). Te chromatographic elution program used was an isocratic one, with 25% A and 75% B, for 25 minutes. Te injected volume of sample and standards was 10 µl. Te DAD detector spectra were recorded between 200 and 600 nm. Te mass spectrometer was equipped with an elec- trospray ionization (ESI) source operated in positive ion mode, and quantifcation was carried out in the multiple reaction monitoring (MRM) mode. Te mass range was between m/z 15 and 1990. Te ion spray temperature was maintained at 250 °C. Te drying gas fow rate was 10 L/min. Evaluation of A26∆sfp and A26 EPS and sodium alginate water holding capacity (WHC). Bacterial cultures were grown and harvested, and polysaccharides isolated, as described above. 600 mg of A26∆sfp and A26 EPS, sodium alginate (Sigma) and levan (Sigma) were mixed with 100 g sand and determined at diferent wetting cycles for 24 h. Te sand and biopolymer mixture was then allowed to drain for 30 min. and the weight afer saturation was recorded. Following the wet weight estimation the mixture was dried in an oven, cooled in a desiccator and weighed again. Each experiment was carried out in triplicate. WHC= gain in weight at saturation point/dry weight of soil × 100.

Data confrmation and validation. To ensure reproducibility, greenhouse experiments were performed in four replicates each containing twenty plants. Twenty biological replicates of each D-glucuronate detection experiment were performed. For the wheat grain assay three biological replicates were performed. Replicated data were studied for normal distribution and analyzed by Unscrambler X15.1 and MiniTab17. One way analysis of variance (ANOVA) and a post-hoc LSD test was used to identify treatments that were signifcantly diferent from controls at p ≤ 0.05. Linear regressions (Unscrambler X15.1) were used to determine the relationships between antagonistic parameters and D-glucuronic acid content. Results P. polymyxa A26∆sfp antagonism against F. graminearum by kernel assay. Te experiment was performed following our former results on A26∆sfp F. graminearum wheat grain assay where grains inoculated with A26∆sfp showed limited pathogen antagonism12. Two initial A26∆sfp inoculum densities, 106 and 108 cells/ ml, were used in the assay. Visual inspection of wheat grains over the experimental period revealed F. gramine- arum mycelial overgrowth in the control treatments. While limited mycelial growth was observed with A26∆sfp at 106 initial inoculum density, no fungal mycelial growth was observed at 108 A26Sfp inoculum density (Fig. 1A and Table S1). Te wheat grain assay was carefully performed under axenic conditions in order to avoid contami- nants from outside. F. graminearum growth randomly confrmed by PCR assays followed by QubitTM Fluorometer quantifcation revealed up to 0.3 ng/µl quantities of the pathogen DNA versus more than 10 ng/µl of the F. gramin- earum control treatment in the case of high initial inoculum (108). Similarly to the assay performed by us earlier12, F. graminearum was detected around 3 ng/µl at low inoculum density (106). Te results show that the A26 mutant, defcient in nonribosomal compound synthesis, is still capable of efciently antagonising the FHB-causing path- ogen F. graminearum in the kernel assay when provided at 108 inoculum density (Fig. 2A and Table S1). Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 4 www.nature.com/scientificreports/

Figure 2. Fusarium graminearum antagonism in kernel assay. (A) Antagonistic activity of Paenibacillus polymyxa A26∆sfp at two inoculum densities, 106 and 108 cells/ml; (B) Antagonistic activity of P. polymyxa A26∆sfp and A26 extracellular polysaccharides (EPS) 15 µg/ml afer 10 days incubation. See Material and Methods.

Antagonistic activity of P. polymyxa A26∆sfp and P. polymyxa A26 in the greenhouse assay. In order to confrm the results with A26∆sfp in reference to its wild-type A26 antagonistic abilities in the wheat grain assay, the experiment was performed in more natural settings, in the greenhouse. Te winter wheat Stava seeds were vernalized and grown by standard procedures until the fowering stage. Ten the wheat heads were sprayed with A26∆sfp and A26 at the two inoculum densities used in the wheat grain assay (106 and 108). Anthesis is the stage of the greatest susceptibility to FHB, as anthers are the common entry route into the plant. Hence, one week afer the antagonist treatment (end of fowering stage) 30 µl of macroconidia suspension was used to inocu- late a single central foret on each wheat head (four replicates each containing twenty plants). Wheat heads were harvested 21 days afer the inoculation at the fully ripe stage and evaluated for disease incidence, disease severity and 100-kernel weight. A26 efciently antagonises F. graminearum and the efect is not dependent on initial inoculum density (Figs 3–5 and Table S1). As shown by the disease incidence (DI), disease severity (DS) and 100- kernel weight (100-kw) (53, 10, and 3.5 respectively) the F. graminearum pathogen was equally well antagonised at both A26 inoculum densities (Figs 3–5 and Table S1). While highly similar antagonistic parameters of DI, DS and 100-kw (53, 10 and 3.6 respectively) were scored for A26∆sfp at the higher inoculum density (108), up to 25% reduced antagonistic ability was observed at 106 density (78, 30 and 2.9 respectively) (Figs 3–5 and Table S1). No disease symptoms were scored on plants afer sole A26 treatment or afer A26Sfp treatment, at either of the two inoculum densities, or afer water treatment (Table S1). 100-kw of the efcient pathogen treatments was highly similar to the kernel weight of control treatments with sole A26 or with A26Sfp or water (3.4 ± 0.6 g).

Wheat head bioflm assessment. In order to follow the performance of A26∆sfp and its wild type A26, wheat heads were fushed with water immediately before the pathogen treatment in order to collect the bio- flms developed afer the A26 or A26∆sfp inoculations. Te bioflm bacteria were re-isolated by selective plating, confrmed by PCR as described13, quantifed, supernatant polysaccharides isolated and D-glucuronate content recorded (Table 2). A26∆sfp and its wild type A26 (103 per ml) were re-isolated in the wheat head bioflms (Table 2). About 30% higher EPS content was detected in A26∆sfp bioflms in comparison to its wild type A26 strain at both initial inoculum densities (Table 2). While neither the EPS content of A26∆sfp nor that of A26 var- ied signifcantly according to the initial inoculum density, the D-GA content of the mutant was about 40% higher at 108 inoculum density. Te D-GA comprised 3.5% of its EPS at 106 inoculum density while the content was 6% at 108 inoculum density (Table 2 and Fig. 6). Te D-GA content in A26 was signifcantly lower than in A26∆sfp. i.e. 3% of A26 EPS (Table 2 and Fig. 6). Te D-GA content correlated with A26∆sfp antagonistic activity (r > 0.78 p < 0.01). Additionally, the EPS titre and its D-GA content in the native bioflm layers of the control plant spikes were evaluated (Fig. 6 and Table 2). Te native bioflms of all the control spikes comprised around 6% D-GA (Table 2 and Fig. 6).

P. polyFig.myxa A26∆sfp and P. polymyxa A26 polysaccharide fragment assessment. UHPLC-ESI-MS under positive ionization was used to determine fragment ions. Figure 7 and Table 3 depict the Q1 mass spectra of the various major chromatographic fractions (Supplementary Information). In positive ion mode, no protonated molecules [M + H]+ were detected, but we clearly identifed ammonium-adducted mole- + cules [M + NH4] , including those at m/z ratios of 191, 388, 567, and 722. Te analyses demonstrated that A26 and A26Sfp bioflm EPS at low and high inoculum densities comprised monomers (176 Da), dimers (352 Da), trimers (528 Da) and tetramers (704 Da). Tere were no signifcant quantities of oligomers larger than tetramers.

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 5 www.nature.com/scientificreports/

Figure 3. Fusarium Head Blight disease incidence (DI) in greenhouse assay. Boxplot of DI scores of wheat plants afer Paenibacillus polymyxa A26 and P. polymyxa A26∆sfp treatments at two inoculum densities, 106 (Low) and 108 (High) cells/ml. ANOVA univariate analysis was performed and post-hoc LSD tests used to identify treatments signifcantly diferent from pathogen control (p < 0.05). Diferent letters indicate statistically signifcant diferences. Error bars represent standard deviations.

Figure 4. Fusarium Head Blight disease severity (DS) in greenhouse assay. Boxplot of DS scores afer Paenibacillus polymyxa A26 and P. polymyxa A26∆sfp treatments at two inoculum densities, 106 (Low) and 108 (High) cells/ml. ANOVA univariate analysis was performed and post hoc LSD test used to identify treatments signifcantly diferent from pathogen control (p < 0.05). Diferent letters indicate statistically signifcant diferences. Error bars represent standard deviations.

Bacterial population EPS titre EPS D-GA F. graminearum log CFU/ml1 (µg/ml) content (%) D-GA (µg/ml) DNA titre (ng/µl)2 Control 1.19 ± 0.11a1* 0.8 ± 0.16a 6 0.05 ± 0.01a 13 ± 2.6 Wheat spike wash Wheat spike wash A26 (Low)3 3.12 ± 0.12b 10 ± 2b 3 0.3 ± 0.06b <0.3 A26 (High) 3.09 ± 0.13b 10 ± 1.5b 3 0.3 ± 0.03b <0.3 Wheat spike wash A26Sfp (Low) 3.02 ± 0.10b 14 ± 2.6c 3.5 0.5 ± 0.1c 4.5 ± 0.8 A26Sfp (High) 3.02 ± 0.09b 15 ± 2.4c 6 0.9 ± 0.05d <0.3 ½ TSB cultures ½ TSB cultures A26 8.89 ± 0.16 10.05 ± 2.23 3 0.3 ± 0.05 ND A26Sfp 8.69 ± 0.11 15.02 ± 2.57 6 0.9 ± 0.09 ND Sodium alginate 15 82 8.3 ± 0.83 ND Levan 15 0 ND ND

Table 2. EPS assessment in relation to bacterial and fungal growth. 1Te bacteria were re-isolated by selective plating and confrmed by PCR13. 1*Counting of bacterial colony forming units was performed on ½ TSB plates 2F. graminearum DNA was isolated amplifed and quantifed using the Invitrogen Qubit 4 Fluorometer as described in Material and Methods. 3Two inoculum densities of A26 and A26Sfp ™(Low) −™106 and (High) 108 cells per ml. Diferent letters indicate statistically signifcant diferences (p < 0.05). ND- not detected.

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 6 www.nature.com/scientificreports/

Figure 5. 100 kernel weight (100-kw) in greenhouse assay. 100- kw afer Paenibacillus polymyxa A26 and P. polymyxa A26∆sfp treatment at two inoculum densities, 106 (Low) and 108 (High) cells/ml. ANOVA univariate analysis was performed and post-hoc LSD tests used to control (p < 0.05). Diferent letters indicate statistically signifcant diferences. Error bars represent standard deviations.

Figure 6. D- glucuronate (D-GA) content in wheat axis bioflms. D-GA content of Paenibacillus polymyxa A26 and A26∆sfp wheat axis bioflms at two inoculum densities, 106 (Low) and 108 (High) cells/ml. ANOVA univariate analysis was performed and post-hoc LSD test used to identify treatments signifcantly diferent from water control (p < 0.05). Diferent letters indicate statistically signifcant diferences. Error bars represent standard deviations.

Antagonistic activity of P. polymyxa A26∆sfp and A26 polysaccharides in the kernel assay. Te greenhouse A26∆sfp treatment at the high inoculum density resulted in signifcant increases in spike bioflm EPS D-GA. Tis correlated with the treatment antagonistic activity. Hence, further studies were performed to study the efect using EPS isolated from A26∆sfp and A26 fask cultures, as well as commercial sodium alginate and levan as reference substrates with contrasting uronate content (Table 2). Te results show that 15 µg/ml EPS treatments of both A26∆sfp and A26 efciently antagonise the F. graminearum pathogen (Fig. 2B and Table S1). While the levan treatment revealed fungal overgrowth comparable to that in the F. graminearum control treat- ment, no pathogen was visually detected in the assay with commercial sodium alginate. F. graminearum growth, randomly confrmed by PCR assays followed by fuorometric quantifcation, revealed up to 0.3 ng/µl quantities of the pathogen DNA in A26∆sfp and A26 EPS and alginate assay versus more than 10 ng/µl with the levan and F. graminearum control treatment (Table S1).

P. polymyxa A26∆sfp and A26 polysaccharide water holding capacity (WHC). Addition of 0.6% (vol) biopolymers signifcantly increased WHC of sand. Te A26 and A26Sfp enhanced the WHC about twice. Te result is comparable to the enhancement by commercial sodium alginate WHC (Table 4). Levan treatment did not improve the sand WHC (Table 4). Discussion Here we re-evaluated the modes of antagonism of the P. polymyxa wild type A26 and its mutant A26∆sfp against F. graminearum (Figs 2–5). While A26 antagonizes efectively at both low (106 cells/ml) and high (108 cells/ ml) densities of inocula used in the study, A26∆sfp F. graminearum antagonism at 108 inoculum density leads to up to 25% improved antagonistic ability, reaching the ability of the wild type to antagonize F. graminearum

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 7 www.nature.com/scientificreports/

Figure 7. Typical examples of mass spectra recorded for diverse fractions of Paenibacillus polymyxa A26 and A26∆sfp wheat spike bioflms at low and high inoculum densities. Note the data presented also in Table 3 (A-I, B-II, C-III, D-IV, E-V). Aqueous solutions of bioflms (10 μl) were injected into the UHPLC–ESI-MS equipment. See Material and Methods.

(Figs 2–5). At the same time, 103 bacteria of both bacterial strains at both inoculum densities were recovered from the wheat heads one week afer inoculation (Table 2). It is commonly known that even though bacteria are crucial for bioflm formation, their measurement is insufcient to quantify bioflms. Te EPS which form the major part of the microbial bioflm matrix19 were detected in the spike bioflm water solutions. Te A26 bioflm contains 1 µg and A26 Sfp 1.5 µg of EPS per ml (Table 2). Te signifcant increase in the mutant bioflm matrix production is in accordance with our former fndings9. Owing to the huge complexity of EPS matrix components, their detailed quantifcation is a challenge. D-glucuronate (D-GA) has previously been suggested as a proxy for

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 8 www.nature.com/scientificreports/

Fraction/RT (min) Oligomer Fragment ions I/2.9 monomer 191; 138; 101; 83; 42 II/3.3 trimer 567; 523; 474; 430; 491; 347; 303; 101; 84; 42 III/3.48 tetramer 784; 696; 652; 608; 564; 520; 476; 179; 138; 117 IV/3.68 dimer 388; 344; 283; 195; 83 V/4.15 tetramer 646; 562; 519; 475; 432; 388; 345; 302; 228; 179; 138; 87; 42

Table 3. Fragment ions of the P. polymyxa A26 and P. polymyxa A26Sfp bioflm spectra.

Control Sodium Alginate Levan A26Sfp A26 WHC 27.6 ± 2.0a2 60.33 ± 2.0b 26.6 ± 2.1a 58.21 ± 3.1b 55.30 ± 2.0b

Table 4. Efect of biopolymer amendment of sand soil WHC1 (%). 1WHC- Water Holding Capacity of sand 0.6%. biopolymer mixture. See Material and Methods. 2Diferent letters indicate statistically signifcant diferences (p < 0.05).

bioflm comprehensive screening and uronic acid is widely determined as representative of myco-polysaccharides in bioflms15. Te assay is based on the efectiveness of chemical bond disruption and digestion of polysaccharides into component monosaccharides15. Signifcant diferences were detected in A26∆sfp D-GA at the two inoculum densities (Fig. 6 and Table 2). Te D-GA comprised 6% of EPS at the 108 inoculum density in comparison to 3.5% at the 106 inoculum density (Fig. 6 and Table 2). Te result correlates well with the ability of the mutant to antagonize F. graminearum (r > 0.78, p < 0.001). Te EPS titre and its D-GA content in the native bioflm layers of the control plant spikes and of the spikes prior to A26 or A26Sfp treatment were evaluated (Fig. 6 and Table 2). It is interesting to note that all the native bioflms on control spikes, similar to A26Sfp high inocula bioflms, com- prised around 6% D-GA. In order to further investigate the role of uronates, two commercially available reference polymers levan and alginic acid were used. Both EPS components can be found in P. polymyxa bioflms20 Te commercial sugars contrast in D-GA content (Table 2). Levan is a polymer of fructose while alginate is comprised of uronates. While the alginate treatment fully antagonized the pathogen, fungal overgrowth was observed in the case of the levan treatment. Te results further confrm the role of uronates regarding the antagonistic efect against F. graminearum. How do we reconcile the outcome that 100 times initial inoculum density diferences result in signifcant increases in the D-GA production of A26∆sfp? Te interesting phenomenon that diferences in initial inoculum density can lead to variable metabolism that cannot be linked to the inocula growth stage, has been reported ear- lier by Jeon et al.21. Te P. polymyxa GBR-1 β-amylase gene was expressed only at a high inoculum density (108 per ml). Te gene was not expressed with the low density inoculum (106)21. Tis phenomenon is certainly connected with complex bioflm biology and requires further study using parallel sequencing and high-resolution micros- copy. However, even though the D-GA composition of the EPS matrix varies between the A26 and A26∆sfp strains, their isolated EPS, applied in surplus, induce resistance to F. graminearum in the wheat kernel assay (Fig. 2B). It is widely recognised that uronic acid moieties afect the physiological and biological properties of poly- saccharides. Te uronates with high content of EPS, e.g. alginate and xanthan, result in increased water holding capacity20 It is believed that uronic acid backbones lead to changes in other sugar backbones, which eventually will result in alteration of their properties and bioactivity22. It is well known that EPS biosynthesis generally involves a very sophisticated network dependent on complex factors23. EPS are combinations of monomers and polymers consisting of sugar compounds connected via glyosidic linkages14,24,25. EPS length, composition and formation vary considerably26–28. Te main factors that infuence EPS interactions include the charges, polymer functional groups, backbone and chain ftness, and the relative concentrations of the constituents14,24,25. Hence EPS are not random co-polymers but vary and occur in combinations dependent on the various C and N sources used for bacterial growth24,25. We were interested in the fragment length of the EPS in A26 and A26∆sfp wheat head bioflms. Te analyses demonstrated that in bioflms from both the strains, the primary compounds were monomers (176 Da), dimers (352 Da), trimers (528 Da) and tetramers (704 Da). Tere were no signifcant quan- tities of oligomers larger than tetramers (Fig. 7, Table 3). Microorganisms in natural environments are subjected to various fuctuations in environmental conditions. It has been suggested that the ecological ‘success’ of EPS depends on their potential to favorably infuence the microorganisms’ adaptation to the environment29. Te EPS matrix serves as the microbial interface with the environment. In addition to environmental adaptation, EPS are also used for social skills communication, com- partmentalization, competence and defense19. As the production of EPS requires copious amounts of energy, their regulatory control is important and there are many levels of EPS synthesis. Diferent subpopulations of eps- producing genes are activated during the diferent stages of bioflm development30. Several EPS molecules have high WHC which can be even as high as 15 times their weight19,20. WHC water accumulation can protect against water stress and was suggested by us as a mechanism of the drought tolerance enhancement of EC, SFS rhizobacterial strains9,10. Te strains from the Hordeum spontaneum rhizosphere are likely to have machinery contributing to the wild cereal progenitors’ adaptation to the unfavorable environment9. In addition to the alle- viation of plant drought stress, the strain WHC contributes to nutrient availability, maintaining water potential

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 9 www.nature.com/scientificreports/

and physiological processes19. In our experiment the D-GA-containing EPS layer with high WHC could provide a physical barrier against F. graminearum pathogens. UHPLC-ESI-MS fragment assessments of the wheat head bioflm EPS afer inoculation with A26 or A26Sfp at low and high inoculum densities showed that all bioflm EPS contain the mono-, di-, tri-, and tetramer fragment mixture (Fig. 7 and Table 3). Yet it is hard to speculate which types of D-GA-containing EPS were produced by the bacterial strains in the assay. It is clear that both A26∆sfp and A26 EPS are capable of absorbing water. Te study in which sand was mixed with the biopolymers showed that A26∆sfp and A26 EPS have high WHC, which is similar to sodium alginate WHC (Table 4). It is interesting to note that the treatment with the commercial low titre (15 µg/ml1) reference substrate alginate resulted in pro- tection against the pathogen in the model system. Te result supports the idea that a biocontrol agent EPS matrix with high WHC is one of the critical factors providing a mechanism against F. graminearum pathogens. Several research groups have considered the strategy of application of pure EPS19. Te work has mostly not resulted in consistent results owing to the complexity of the physical and chemical conditions of the natural environments as well as to the presence of native microbial communities which metabolize the EPS19. On the other hand, it is possible to engineer the EPS production of the native bioflm formers via the nutrient supply14. Here we present a study where the initial observations of the antagonism of the P. polymyxa A26 bioflm to F. graminearum in the kernel assay are examined at a coarser scale in the greenhouse, and at a fner scale in exper- iments with commercial EPS application. We show that A26 bioflm EPS are capable of antagonizing F. gramine- arum and that the uronate content of the EPS is of critical importance in the antagonism. It is clear that fundamental understanding of the genes and mechanisms of real-time bioflm formation is needed to explain the performance of the P. polymyxa native isolate A26 and its mutant A26∆sfp. Previous studies have shown that the multifaceted information about the bacterial bioflm cannot simply be adopted from studies of their domesticated strains31. It is possible that the A26 uronate-containing EPS matrix is required for efcient antagonism in the cases where the more common mechanisms of antagonism with antibiotic lipopeptides4,9 are not available (as with A26∆sfp). Tis would be in accordance with the fact that surplus EPS was used in the grain assay. How much the EPS produced by wild type P. polymyxa A26 contribute to its F. graminearum antagonism under diferent environmental conditions remains to be elucidated. Te study here indicates that along with the lipopeptides4,9, the bacterial bioflm EPS compounds are capable of antagonizing F. graminearum and that the uronate content of the polysaccharides is of critical importance. Bacterial EPS is a promising class of sustainable biopolymers to meet various industrial/agricultural require- ments. Considering the presence of bacterial natural surface bioflms and that only about 1% of the bacteria can at present be cultured, many EPS are yet to be identifed23,32. Even though the EPS layer is dependent upon the perception of numerous environmental signals from the host and the ecosystem, this would open the new range of application of EPS in integrated pathogen management protecting against stress factors under climate change. Conclusions Paenibacillus polymyxa A26 represents a promising biocontrol agent for addressing the many challenges F. graminearum poses to agricultural crop yield and quality. However, to capitalize on this potential, an improved understanding of the mechanism of action is needed. Tis study demonstrates that the bioflm polysaccharides containing D-GA signifcantly contribute to antagonism and that the initial inoculum density plays a signifcant role in this. References 1. Chen, L., Heng, J., Qin, S. & Bian, K. A comprehensive understanding of the biocontrol potential of velezensis LM2303 against Fusarium head blight. PLoS One 13, e0198560 (2018). 2. Palazzini, J., Torres y Torres Lara, C. & Chulze, T. Biological Control of Fusarium Head Blight of Wheat: From Selection to Formulation. (Springer, Dordrecht 2013). 3. Khan, N. et al. Antifungal activity of Bacillus species against Fusarium and analysis of the potential mechanisms used in the biocontrol. Frontiers in Microbiol (2018). 4. Mousa, W. K., Shearer, C. R., Limay-Rios, V., Zhou, T. & Raizada, M. N. Bacterial endophytes from wild maize suppress Fusarium graminearum in modern maize and inhibit mycotoxin accumulation. Frontiers in plant science 6, 805 (2015). 5. He, J., Boland, G. J. & Zhou, T. Concurrent selection for microbial suppression of Fusarium graminearum, Fusarium head blight and deoxynivalenol in wheat. J Appl Microbiol 106, 1805–1817 (2009). 6. Branda, S. S., Vik, S., Friedman, L. & Kolter, R. Bioflms: the matrix revisited. Trends Microbiol 13, 20–26 (2005). 7. Davey, M. E. & O’Toole, G. A. Microbial bioflms: from ecology to molecular genetics. Microbiol Mol Biol Rev 64, 847–867 (2000a). 8. Redmile-Gordon, M., Brooks, B., Evershed, R., Goulding, K. & Hirsch, P. R. Measuring the soil-microbial interface: Extraction of extracellular polymeric substances (EPS) from soil bioflms. Soil Biol Biochem 72, 163–171 (2014). 9. Timmusk, S. et al. Sfp- type PPTase inactivation promotes bacterial bioflm formation and ability to enhance plant drought tolerance Frontiers in Microbiol, 1–13 (2015). 10. Nevo, E. “Evolution Canyon,” a potential microscale monitor of global warming across life. Proc Natl Acad Sci USA 109, 2960–2965 (2012). 11. Timmusk, S. et al. Bacterial distribution in the rhizosphere of wild barley under contrasting microclimates. PLoS One 6, 1–8 (2011). 12. Abd El-Daim, I., Haggblom, P., Karlsson, M., Stenstrom, E. & Timmusk, S. Paenibacillus polymyxa A26 Sfp-type PPTase inactivation limits bacterial antagonism against Fusarium graminearum but not of F. culmorum Front. Plant Sci., 1–8 (2015). 13. Timmusk, S. et al. Drought-tolerance of wheat improved by rhizosphere bacteria from harsh environments: enhanced biomass production and reduced emissions of stress volatiles. PloS ONE 1–13 (2014). 14. Rutering, M., Schmid, J., Ruhmann, B., Schilling, M. & Sieber, V. Controlled production of polysaccharides-exploiting nutrient supply for levan and heteropolysaccharide formation in Paenibacillus sp. Carbohydr Polym 148, 326–334 (2016). 15. Mojica, K., Elsey, D. & Cooney, M. J. Quantitative analysis of bioflm EPS uronic acid content. J Microbiol Methods 71, 61–65 (2007). 16. Timmusk, S., Paalme, V., Lagercrantz, U. & Nevo, E. Detection and quantifcation of Paenibacillus polymyxa in the rhizosphere of wild barley (Hordeum spontaneum) with real-time PCR. J Appl Microbiol 107, 736–745 (2009). 17. Nygren, C. M. et al. Growth on nitrate and occurrence of nitrate reductase-encoding genes in a phylogenetically diverse range of ectomycorrhizal fungi. Te New phytologist 180, 875–889 (2008). 18. Gardes, M. & Bruns, T. D. ITS primers with enhanced specifcity for basidiomycetes–application to the identifcation of mycorrhizae and rusts. Mol Ecol 2, 113–118 (1993).

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 10 www.nature.com/scientificreports/

19. Costa, O. Y. A., Raaijmakers, J. M. & Kuramae, E. E. Microbial Extracellular Polymeric Substances: Ecological Function and Impact on Soil Aggregation. Frontiers in microbiology 9, 1636 (2018). 20. Liang, T. W. & Wang, S. L. Recent advances in exopolysaccharides from Paenibacillus spp.: production, isolation, structure, and bioactivities. Mar Drugs 13, 1847–1863 (2015). 21. Jeon, Y. H., Kim, S. G., Hwang, I. & Kim, Y. H. Efects of initial inoculation density of Paenibacillus polymyxa on colony formation and starch-hydrolytic activity in relation to root rot in ginseng. J Appl Microbiol 109, 461–470 (2010). 22. Xu, X. et al. Unsaturated guluronate oligosaccharide enhances the antibacterial activities of macrophages. FASEB J 28, 2645–2654 (2014). 23. Schmid, J., Sieber, V. & Rehm, B. Bacterial exopolysaccharides: biosynthesis pathways and engineering strategies. Frontiers in microbiology 6, 496 (2015). 24. Ruhmann, B., Schmid, J. & Sieber, V. High throughput exopolysaccharide screening platform: from strain cultivation to monosaccharide composition and carbohydrate fngerprinting in one day. Carbohydr Polym 122, 212–220 (2015). 25. Ruhmann, B., Schmid, J. & Sieber, V. Methods to identify the unexplored diversity of microbial exopolysaccharides. Frontiers in microbiology 6, 565 (2015). 26. Yegorenkova, I. V., Tregubova, K. V., Burygin, G. L., Matora, L. Y. & Ignatov, V. V. Assessing the efcacy of co-inoculation of wheat seedlings with the associative bacteria Paenibacillus polymyxa 1465 and Azospirillum brasilense Sp245. Can J Microbiol 62, 279–285 (2016). 27. Yegorenkova, I. V., Tregubova, K. V., Matora, L. Y., Burygin, G. L. & Ignatov, V. V. Bioflm formation by Paenibacillus polymyxa strains difering in the production and rheological properties of their exopolysaccharides. Curr Microbiol 62, 1554–1559 (2011). 28. Yegorenkova, I. V., Tregubova, K. V., Matora, L. Y., Burygin, G. L. & Ignatov, V. V. Use of ELISA with antiexopolysaccharide antibodies to evaluate wheat-root colonization by the rhizobacterium Paenibacillus polymyxa. Curr Microbiol 61, 376–380 (2010). 29. Geoghegan, M. et al. Te polymer physics and chemistry of microbial cell attachment and adhesion. Faraday Discuss 139, 85–103; discussion 105–128, 419–120 (2008). 30. Chai, Y., Chu, F., Kolter, R. & Losick, R. Bistability and bioflm formation in . Mol Microbiol 67, 254–263 (2008). 31. Aguilar, C., Vlamakis, H., Losick, R. & Kolter, R. Tinking about Bacillus subtilis as a multicellular organism. Curr Opin Microbiol 10, 638–643 (2007). 32. Stewart, E. J. Growing unculturable bacteria. J Bacteriol 194, 4151–4160 (2012). Acknowledgements Te work was performed with fnancial support from the Carl Tryggers Stifelse för Vetenskaplig Forskning, Swedish Research Council 2014-04035, FORMAS 222-2014-1326 and the European Commission and the Romanian Government (POSCCE 621/2014). We are indebted to Dr. D. Clapham for critically reading the manuscript. Author Contributions All Authors made equal contribution to the planning and development of experiments to investigate the original idea by S.T. D.C. and L.C. contributed most to the chemical aspects. S.T., T.T., E.N. and L.B. contributed most to the biological aspects. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-37718-w. Competing Interests: Te authors declare no competing interests. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© Te Author(s) 2019

Scientific Reports | (2019)9:662 | https://doi.org/10.1038/s41598-018-37718-w 11